Adjustments for under-reporting in alcohol surveys have been used in epidemiological and policy studies which assume that all drinkers underestimate their consumption equally. This study aims to describe a method of estimating how under-reporting of alcohol consumption might vary by age, gender and consumption level. The Canadian Alcohol and Drug Use Monitoring Survey (CADUMS) 2008-10 (n = 43 371) asks about beverage-specific 'yesterday' consumption (BSY) and quantity-frequency (QF). Observed drinking frequencies for different age and gender groups were calculated from BSY and used to correct values of F in QF. Beverage-specific correction factors for quantity (Q) were calculated by comparing consumption estimated from BSY with sales data. Drinking frequency was underestimated by males (Z = 24.62, P < 0.001) and females (Z = 17.46, P < 0.001) in the QF as assessed by comparing with frequency and quantity of yesterday drinking. Spirits consumption was underestimated by 65.94% compared with sales data, wine by 38.35% and beer by 49.02%. After adjusting Q and F values accordingly, regression analysis found alcohol consumption to be underestimated significantly more by younger drinkers (e.g. 82.9 ± 1.19% for underage drinkers versus 70.38 ± 1.54% for those 65+, P < 0.001) and by low-risk more than high-risk drinkers (76.25 ± 0.34% versus 49.22 ± 3.01%, P < 0.001). Under-reporting did not differ by gender. Alcohol consumption surveys can use the beverage-specific 'yesterday method' to correct for under-reporting of consumption among subgroups. Alcohol consumption among Canadians appears to be under-reported to an equal degree by men and women. Younger drinkers under-report alcohol consumption to a greater degree than do older drinkers, while low-risk drinkers underestimate more than do medium and high-risk drinkers. © 2014 Society for the Study of Addiction.